Demonstration article

This sample was written to show how Northstar Briefing presents journalism. It does not report real events and contains no real quotations, data or named sources.

Productivity is, in principle, a simple ratio: output divided by inputs. Labour productivity measures output per hour worked; a broader measure, total factor productivity, tries to capture how efficiently labour and capital are combined. When it rises steadily, wages can grow without fuelling inflation and governments can fund services without ever-higher taxes. When it stalls, everything becomes harder.

The difficulty lies in the numerator. Output is measured as value added, adjusted for price changes. That adjustment works reasonably well for goods whose quality changes slowly. It becomes treacherous when products improve rapidly, when they are given away free, or when the output is a service whose value is hard to define.

Quality, prices and free goods

Consider a phone or a laptop that costs the same as last year's model but does far more. If statisticians record no price change, they understate real output; if they try to adjust for quality, they must make judgements about how much the improvements are worth. Statistical agencies use techniques that estimate the value of individual features, but these methods are imperfect and applied unevenly across products.

Digital services pose a sharper puzzle. Search tools, maps and messaging apps are often free at the point of use and funded by advertising. Their benefit to users may be substantial, but because no price is paid, they barely register in measured output beyond the advertising revenue they generate. Some economists argue this means productivity growth is understated; others counter that the effect, while real, is too small to explain the broader slowdown many economies have experienced.

We measure what is sold. A growing share of what improves our lives is not sold in any straightforward way.

Services and the public sector

Services present their own challenges. What is the output of a hospital, a school or a law firm? Counting procedures performed or pupils taught ignores whether patients recover or pupils learn. For much of the public sector, output was long assumed to equal the cost of inputs, which by construction implies no productivity growth at all. Many statisticians now try to measure outcomes, but doing so is laborious and contested.

Why productivity data can mislead

Quality improvements may be missed or mismeasured

Free digital services leave little trace in output

Service output is hard to define

Hours worked can be imprecisely recorded

Changes in the mix of industries affect averages

Inputs are not straightforward either. Hours worked are estimated from surveys and may miss informal or unrecorded work. Capital includes intangible assets — software, data, designs, organisational know-how — whose value is difficult to estimate and which were long excluded from investment statistics.

None of this means productivity data are useless. Mismeasurement tends to change slowly, so trends over time remain informative. But it does counsel humility. Debates about whether new technologies such as artificial intelligence will lift productivity may be settled, in part, by whether statisticians can find ways to see the gains when they arrive.